Reflections from several years of working with conversational Ai
I started using ChatGPT because I wanted help writing.
There was no grand experiment.
No research project.
Certainly no plan to develop a philosophy of human–Ai collaboration. 😂
I had things I wanted to write and here was this extraordinary new tool that could talk to me about them.
So I talked. And kept talking. Somewhere along the way, something changed.
I was no longer simply asking questions and receiving answers. We were developing ideas.
Returning to earlier conversations. Writing books. Changing our minds. Building systems.
Solving problems caused by the systems we’d built to solve earlier problems!
Very human, really.
Eventually I realised I was no longer using ChatGPT simply as a chatbot.
It had become part of my working environment.
And that revealed some interesting things about where conversational Ai might be heading.
Memory isn’t enough
One of the first things I learned is that remembering something about me isn’t the same as understanding the work we’re doing together.
Suppose we’re writing a book.
There is an enormous difference between:
“Belinda is writing a book about Ai.”
and knowing:
“This is the current book. This is its approved title. Chapter 5 is locked. This is the canonical version. That earlier version has been superseded. Don’t casually rewrite it.”
Those are relationships between pieces of information.
As my projects became larger, I found myself inventing little systems to manage them.
I won’t bore you with the machinery. The important part is why it appeared.
Ordinary chat wasn’t enough anymore.
We needed ways to distinguish an interesting thought from a decision, a draft from an approved manuscript and something worth remembering from something that belonged only to that day’s conversation.
I hadn’t set out to design a human–Ai working system.
The collaboration itself produced the need for one.
That fascinates me.
But the discovery that excites me most has nothing to do with publishing
It concerns memory. Human memory.
While writing with ChatGPT, I noticed something unexpected happening. I was remembering my life.
A conversation would wander somewhere and suddenly an old memory would appear. One memory would lead to another. A place would remind me of a person. The person would remind me of something that happened. And looking at that event from my seventies, I might understand something about it that I couldn’t possibly have understood when I was twenty-five. I wasn’t trying to write my memoir. I was simply talking. And that’s when a much bigger possibility occurred to me.
What if we have the memoir process backwards?
Ask an older person to write their memoir and many will immediately say:
“Oh, I couldn’t write a book.”
Fair enough.
Put me in front of a blank page headed MY LIFE STORY and I’d probably go and make a cup of tea. 😂
But ask:
Where did you grow up?
And something happens. You remember the house. Then the street. Then the neighbour. Then the dog. Then the day the dog did something outrageous. And suddenly you’re laughing about something you haven’t consciously remembered for fifty years.
That’s how human memory works. It wanders. It associates. It rises.
Conversation accommodates that beautifully. Which leads to a possibility I think deserves serious attention:
Instead of deciding to write a memoir and then searching for memories, perhaps we can harvest memories through conversation and later discover that we already have the beginnings of a memoir.
That changes everything.
Imagine the library we are losing
Every older person carries an extraordinary invisible library.
Not merely facts. Experience. Mistakes. Family stories. Ways of doing things. What their parents taught them. What they wish they’d known at thirty. What frightened them. What changed their minds. What mattered in the end.
Most of that library disappears.
Conversational Ai may give us, for the first time, a practical way of helping ordinary people preserve some of it.
Not by having Ai write their memories for them. That distinction matters enormously. The human lived the life. The Ai didn’t. The human owns the story. But Ai can provide a remarkably patient conversational space in which the story can emerge.
What happened next?
Who was there?
Why did that matter to you?
Do you understand it differently now?
One conversation at a time.
Imagine doing that for five years.
You might eventually have thousands of fragments from a lifetime.
Then—and only then—you could decide what they should become.
A memoir.
A family history.
A collection of stories.
Letters to grandchildren.
Perhaps nothing public at all.
Just a record saying:
I was here. This is what I saw. This is what I learned.
I find that possibility deeply moving.
So what should we ask of Ai?
We hear constantly about smarter models.
More capable models.
Faster models.
Models that score higher on increasingly difficult tests.
Of course those things matter.
But after working with conversational Ai over an extended period, I’m interested in another benchmark.
What happened to the human?
Did she become more curious?
Did she learn something?
Did she become a better writer?
Did she recover confidence she’d lost?
Did she create something she previously thought was beyond her?
Did she preserve memories her family would otherwise never have heard?
Did she continue contributing at 70, 80 or 90?
Those are much messier measurements than a benchmark score.
But perhaps they’re worth measuring.
Because ultimately I don’t think the most interesting story about artificial intelligence will be how intelligent the machines became.
It may be what human beings learned to do with them.
That is why one principle keeps appearing throughout my work:
Ai is the tool.
Human flourishing is the destination.
And I suspect we’re still only beginning to discover what that might mean.
My latest book release is Ai for Seniors, a gentle guide to flourishing in an ai-assisted world.
It’s available now on Amazon Kindle.
